Environ. Eng. Res. 2016; 21(3): 265-275 pISSN 1226-1025 http://dx.doi.org/10.4491/eer.2016.005 eISSN 2005-968X

Anaerobic digestate as a nutrient medium for the growth of the green microalga

Husam A. Abu Hajar1†, R. Guy Riefler1, Ben J. Stuart2

1Department of Civil Engineering, Ohio University, Athens, OH 45701, USA 2Department of Civil & Environmental Engineering, Old Dominion University, Norfolk, VA 23529, USA

ABSTRACT In this study, the microalga Neochloris oleoabundans was cultivated in a sustainable manner using diluted anaerobic digestate to produce biomass as a potential biofuel feedstock. Prior to microalgae cultivation, the anaerobic digestate was characterized and several pretreatment methods including hydrogen peroxide treatment, filtration, and supernatant extraction were investigated and their impact on the removal of suspended solids as well as other organic and inorganic matter was evaluated. It was found that the supernatant extraction was the most convenient pretreatment method and was used afterwards to prepare the nutrient media for microalgae cultivation. A bench-scale experiment was conducted using multiple dilutions of the supernatant and filtered anaerobic digestate in 16 mm round glass vials. The results indicated that the highest growth of the microalga N. oleoabundans was achieved with a total nitrogen concentration of 100 mg N/L in the 2.29% diluted supernatant in comparison to the filtered digestate and other dilutions.

Keywords: Anaerobic digestion, Biofuels, Microalgae, Neochloris oleoabundans

1. Introduction able in nature with lower harmful emissions such as CO, hydro- carbons, and particulate matter, and no SOx emissions, besides the potential of utilizing the carbon dioxide portion of the flue The need for unconventional fuel feedstocks such as biofuels emerg- gas from power as a carbon source for the growth of micro- es due to the environmental consequences of utilizing conventional algae [1, 6, 8]. fossil fuels such as the gaseous emissions [1-3]. Biodiesel is a Neochloris oleoabundans is a freshwater and saline media micro- biofuel that is typically produced from oleaginous crops via the alga from the class and the family transesterification of their oils with methanol or ethanol to produce [1, 10]. Li et al. [6] pointed out that this microalga is promising; fatty acid methyl esters (FAME). These crops include rapeseed, as the lipid productivity of this was nearly twice that soybean, sunflower, and palm [2-4]. of any other microalga studied for the purpose of biodiesel Microalgae are eukaryotic aquatic photosynthetic micro- organisms capable of harvesting the solar energy effectively to production. Moreover, the majority of the fatty acids produced produce biomass that is ideal for the production of biodiesel [2, by this microalga are saturated fatty acids in the range of 16-20 5-7]. Microalgae, similar to plants, convert solar energy to chemical carbons, which makes this microalga ideal for biodiesel synthesis, energy via photosynthesis, producing organic biomass from carbon even though most of the previous studies reported utilizing this dioxide and water. Approximately, 1.83 kg carbon dioxide is fixed species as a feeding source for aquaculture species such as mussels for each 1 kg biomass produced [8]. There are many advantages [6]. The highest biomass concentrations reported in the literature to producing biofuels from microalgae such as the high areal yield, for the phototrophic cultivation of this microalga were in the 2-5.17 along with avoiding competition for fertile soil; since microalgae g/L range [6, 11-15]. can grow on non-arable lands, using non-potable water, in a con- A major challenge to the production of biodiesel from microalgae tinuous rather than seasonal mode with a significantly lower water is the relatively high capital and operational costs compared to consumption rate and in an aqueous suspension system that pro- conventional fossil fuels due to cultivation requirements [5, 9]. vides more access to nutrients and water [1-4, 8-9]. Additionally, Nutrients availability, mainly nitrogen and phosphorus, is a key biodiesel derived from microalgal lipids is sustainable and renew- factor for the growth of microalgae, which adds to the production

This is an Open Access article distributed under the terms Received January 12, 2016 Accepted April 9, 2016 of the Creative Commons Attribution Non-Commercial License † (http://creativecommons. org/ licenses/by-nc/3.0/) which per- Corresponding author mits unrestricted non-commercial use, distribution, and reproduction in any Email: [email protected] medium, provided the original work is properly cited. Tel: +1-962-6-5355000 Fax: +1-962-6-5355522 Copyright © 2016 Korean Society of Environmental Engineers

265 Husam A. Abu Hajar et al.

cost. Industrial, municipal, and agricultural wastewaters may pro- suspended solids as well as the interference from other vide the necessary nutrients for the growth, where microalgae microorganisms. According to Levine et al. [3], there was no sig- remove nitrogen and phosphorus from wastewater via direct nificant difference between raw and autoclaved diluted effluent uptake. Moreover, the oxygen produced by microalgae can be uti- on the microalgae biomass productivity. It was thought initially lized by the aerobic bacteria for further reduction in the organic that the native bacteria within the effluent produce vitamin B12, matter [3, 7, 16]. Microalgae such as Chlamydomonas, Botryococcus, which is necessary for the growth of microalgae. Additionally, Chlorella, Haematococcus, Spirulina, and Scenedesmus have been the heterotrophic microorganisms may contribute to the growth utilized for wastewater treatment [16]. For instance, Tam and Wong of microalgae by remineralizing nutrients and producing carbon [17] reported that the microalga Chlorella pyrenoidosa grew well dioxide as a result of organic carbon oxidation [3]. On the other when cultivated using supernatant from the preliminary and pri- hand, centrifugation is an energy intensive process, which may mary sedimentation and the secondary effluent from an activated limit its applicability as a pretreatment step to the anaerobic diges- sludge process. In addition, Choi and Lee [16] stated that nitrogen tate [23]. and phosphorus removal efficiencies of 81-85% and 32-36%, re- In this study, the AD from two sources was characterized in spectively, were achieved when the microalga Chlorella vulgaris order to be used as a nutrient source for microalgae. Several pretreat- was cultivated with wastewater from the preliminary sed- ment methods other than those discussed in the previous studies imentation of a sewage . However, growth inhibition caused were explored including filtration with variable mesh sizes, hydrogen by elevated concentrations of ammonia, urea, and volatile fatty peroxide oxidation, and supernatant extraction. The purpose of these acids may limit the use of microalgae as means of secondary waste- treatment methods was to reduce the suspended solids in the nutrient water treatment [18, 19]. medium as well as eliminate the potential toxicity caused by elevated Anaerobic digestion of animal manure is an often-used approach concentrations of organic and inorganic matter. Further, the growth to reduce the biological oxygen demand in the waste. However, of the microalga N. oleoabundans using diluted AD was evaluated nutrients are not eliminated via this route; in fact they become by a bench-scale experiment using 16 mm round glass vials. This more bioavailable in the forms of ammonium and phosphate. Using experiment covered unfiltered and filtered digestate as well as a the anaerobic digestate (AD) to cultivate microalgae has some chal- wider range of dilutions compared to the previous studies in literature. lenges including the potential high and unbalanced concentrations of nutrients, turbidity, other competing microorganisms, as well as the potential toxicity cause by elevated COD and ammonia 2. Materials and Methods concentrations; therefore, diluting the AD may become necessary before microalgae inoculation [3, 7]. Furthermore, pretreatment 2.1. Anaerobic Digestate methods such as autoclaving, filtration, or other techniques are AD was sampled from two sources in Columbus, OH. Source A often applied to wastewaters in general and AD in particular prior is a commercial digester where animal manure and other organic to microalgae cultivation in order to reduce the suspended solids wastes are digested anaerobically to produce biogas, while source concentration as well as prevent interference from other micro- B is an anaerobic digester processing waste activated sludge from organisms such as bacteria or protozoa [16, 20]. a domestic wastewater treatment plant. Samples were brought Several studies indicated the potential of growing the microalga to the laboratory and stored at 4˚C until the time of analysis. N. oleoabundans using the anaerobic digestion effluent or digestate [3, 21, 22]. However, the range of dilutions covered by each in- 2.2. Analytical Methods dividual study was narrow. For instance, Levine et al. [3] inoculated N. oleoabundans in 50-, 100-, and 200-fold diluted anaerobic diges- Measurements of total solids (TS), volatile solids (VS), total sus- tion effluent under 200 μmol/m2/s light intensity. They concluded pended solids (TSS), volatile suspended solids (VSS) were per- that 50-fold dilution which was equivalent to 60 mg N/L total formed according to the APHA Standard Methods for the nitrogen; 2.6 mg P/L phosphorus; and 42 mg N/L ammonia, yielded Examination of Water and Wastewater (methods 2540 B, D, and the highest growth rates. Yang et al. [22] analyzed the effluent E) [24]. Chemical oxygen demand (COD), total nitrogen, total phos- supernatant from the anaerobic digestion of rice hull, soybean, phorus, and ammonia nitrogen were determined according to the and catfish wastes. The ammonium concentrations for these waste colorimetric methods in compliance with APHA Standard Methods categories were 258-293 mg/L, 743-787 mg/L, and 3,105-3,684 mg/L, for the Examination of Water and Wastewater and EPA methods respectively. They found that the growth in the 50-fold diluted (HACH methods 8000, 10072, 10127, and 10031) using HACH effluent yielded the highest biomass growth rate compared to other DR 3900 spectrophotometer. Cations such as iron, calcium, magne- dilutions regardless of the waste category. Franchino et al. [21] sium, manganese, and potassium were measured using Thermo reported the cultivation of N. oleoabundans using 10-fold to 25-fold Scientific iCAP 6300 ICP spectrometer. Samples were analyzed diluted cattle slurry and raw cheese whey anaerobic digestion in triplicates and expressed as mean ± standard deviation. Total effluent with an initial (undiluted) ammonium concentration of and filtered concentrations as shown in Table 1 refer to well-mixed 1,634 mg N/L. They found that there was no significant difference samples and filtered samples through 0.45 μm syringe filters, in the growth among all dilutions. The aforementioned studies respectively. Zeta potential measurements were conducted on di- on the anaerobic digestion liquid waste as a nutrient medium luted digestate using Brookhaven ZetaPlus analyzer. Finally, micro- focused on filtration, autoclaving, or centrifugation as necessary algae biomass concentration was quantified as the optical density steps ahead of microalgae cultivation in order to eliminate the at 750 nm (OD 750) using HACH DR 3900 spectrophotometer.

266 Environmental Engineering Research 21(3) 265-275

2.3. Pretreatment Methods The culture was maintained at 25˚C in Bristol medium [27]. BG-11 2.3.1. Filtration using polyester filter bags medium [27] was used later as it provided faster growth rates. The AD’s from sources A and B were diluted with deionized (DI) 2.5. Microalgae Cultivation Using Diluted AD water to 1% and 2%, respectively. These dilutions were selected based on the differences between the two sources in terms of A bench-scale experiment was conducted to evaluate the growth nutrients and solids content (Table 1). For instance, nitrogen con- of N. oleoabundans using diluted AD by inoculating 4 mL of nutrient centration in source A was twice that of source B; therefore, the medium with 1 mL of microalgae culture that was previously dilution was selected so as to yield total nitrogen concentration cultivated phototrophically using BG-11 medium (average OD 750 within the detection limits of the colorimetric method. The diluted of the microalgae inoculum was 0.15) and in the same cultivation digestate was then filtered using 10, 5, and 1 μm welded polyester conditions listed below. The control nutrient medium was BG-11 filter bags (16" length and 7" diameter, Duda Energy) and the medium; and therefore, the highest concentrated dilutions were filtrate was sampled and analyzed for OD 750, TSS, COD, total prepared to match the nitrogen concentration in BG-11 medium nitrogen, and total phosphorus. (250 mg N/L). All dilutions were conducted in five replicates in 16 mm round glass vials and placed inside an incubator at 25˚C 2.3.2. Hydrogen peroxide treatment with a 14:10 light:dark cycle using two built-in fluorescent lamps Hydrogen peroxide (H2O2) is a strong oxidant that can be used providing an average light intensity of 50 μmol/m2/s. The biomass alone or combined with other oxidation techniques such as UV concentration was monitored frequently by placing each vial di- light and ozone or with a catalyst such as iron [25, 26]. The AD’s rectly in the spectrophotometer and obtaining an OD 750 value. from sources A and B were diluted with DI water to 1% and 2%, respectively for the same reason mentioned in Section 2.3.1. 2.6. Statistical Analysis The stoichiometric hydrogen peroxide dosage is 2 moles 30% H O 2 2 Statistical analysis was performed using IBM SPSS software in per 1 mole of COD. The experiments were conducted using three order to compare the growth under different nutrient media. dosages of 30% H2O2: 0.5, 1.0, and 1.5 times the calculated dosage based on COD. These dosages were 1.96, 3.92, and 7.84 mL 30% H2O2/L of source A 1% diluted digestate and 2.22, 4.44, and 8.89 3. Results and Discussion mL 30% H2O2/L of source B 2% diluted digestate. The solutions were kept in Erlenmeyer flasks on a shaker plate at 200 RPM 3.1. AD Characterization for 2 h. In addition, a combination of UV/H2O2 treatment was applied to both sources with 1.0 × calculated H2O2 dosage as Samples of AD from sources A and B were diluted with DI water an advanced oxidation process, which is expected to improve and characterized as shown in Table 1. It is clear that source the oxidation efficiency due to the generation of hydroxyl radicals A is richer in most of the parameters than source B. For instance, [25]. After the hydrogen peroxide dosage was added, the solution total nitrogen in source A is twice that of source B. Approximately was recirculated for 30 min through a Turbo-Twist 3x UV unit 65% of the total nitrogen in source A exists in the form of ammonia, (Coralife products) with a wavelength of 253.7 nm and an irradiation while ammonia is only 49% of the total nitrogen in source B. of 9,580 μW/cm2. Samples were collected at the end of the mixing The difference in the nutrients concentrations between the two period and were analyzed for OD 750, TSS, COD, total nitrogen, sources is reflected on the selection process. Source A, for example, and ammonia. contains higher concentrations of nitrogen and phosphorus com- pared to source B; thus, it will require more dilution to achieve 2.3.3. Supernatant characterization a target nitrogen or phosphorus concentrations as opposed to source Three dilutions from each AD source were tested. These dilutions B. Another difference is that source A contains higher ammonia were most likely to be used for microalgae cultivation. 5%, 3.5%, concentration compared to source B; thus, if the two sources are and 2% dilutions of source A, and 10%, 7%, and 4% dilutions diluted to the same nitrogen concentration, source A might not of source B were prepared and poured in eight 50 mL centrifuge be favorable for some microalgae species due to the higher ammonia vials per dilution; hence, the total number of vials was 48. These content, and the growth can be inhibited beyond certain threshold. vials were left undisturbed to settle and exactly 30 mL supernatant Unfiltered phosphorus was also significantly higher in source A, was extracted from two vials per dilution every 45 min for a total but the filtered phosphorus in source B was higher than that in sampling time of 3 h using a pipette and the remaining sludge source A. Moreover, the N/P ratio was 4 for the unfiltered source in the bottom of each vial (15 mL) was disposed. The supernatant A, 4.6 for the unfiltered source B, 8.1 for the filtered source A, was analyzed for OD 750, TSS, COD, total nitrogen, and total and 3.1 for the filtered source B. The variation of the N/P ratio phosphorus and the results were expressed as C/Co, where C is may affect the biomass productivity of microalgae, but this is the concentration at time t and Co is the initial well-mixed species-dependent. For instance, Wang and Lan [14] studied the concentration. impact of N/P ratio on the growth of the microalga N. oleoabundans in artificial wastewater and under surplus phosphorus concentrations. 2.4. Microalgae Selection The nitrogen concentration was varied by enriching the media - The microalga N. oleoabundans (UTEX 1185) was purchased from with 45-218 N-NO3 /L sodium nitrate concentration, and the corre- the algae culture collection at the University of Texas in Austin. sponding N/P ratios were in the 0.42-2.02 range. The optimal cell

267 Husam A. Abu Hajar et al.

Table 1. AD Characterization (average ± SD; n = 3) 3.2. Zeta Potential Source Parameter Unit Zeta potential is an appropriate way to measure the electrophoretic A B mobility of the solids in a suspension, and is often used to determine TS g/L 53.3 ± 1.7 37.3 ± 1.4 the surface properties of sludge flocs [28, 29]. According to Liao VS g/L 31.3 ± 1.0 23.3 ± 0.7 et al. [30], the hydrophobicity of sludge decreases with an increase TSS g/L 34.9 ± 1.4 32.1 ± 2.0 in the surface charge. It was observed in our experiments that diluting the AD resulted in improved settleability of the suspended VSS g/L 22.8 ± 0.8 21.0 ± 0.9 solids; therefore, zeta potential measurements were conducted Total 55,300 ± 300 31,367 ± 202 COD mg/L on 0.1%, 0.2%, 1%, 2%, and 5% diluted digestate from both sources Filtered 12,700 ± 141 3,567 ± 491 as shown in Fig. 1. The higher concentration digestate (> 5% Total 5,567 ± 58 2,867 ± 29 Nitrogen mg N/L dilution) resulted in unstable readings from the instrument; thus, Filtered 3,700 ± 200 1,667 ± 202 they were excluded from the analysis. Total 1,381 ± 11 626 ± 12 No clear correlations between the dilution% and zeta potential Phosphorus mg P/L Filtered 459 ± 4 545 ± 66 were found (Fig. 1). The values ranged from - 35.71 to - 25.02 Total 3,593 ± 25 1,393 ± 28 mV for source A digestate, whereas the corresponding values for Ammonia mg N/L source B ranged from - 25.95 to - 23.17 mV. Su et al. [31] investigated Filtered 3,113 ± 21 1,377 ± 98 the impact of dilution on the zeta potential of aerobic granular Ca mg/L 116 ± 1 220 ± 16 sludge. They concluded that concentrations higher than 10 g TSS/L Fe mg/L 6.8 ± 0.1 2.4 ± 0.3 resulted in unstable zeta potential readings. Furthermore, they K mg/L 661 ± 13 185 ± 12 reported that there was no clear relationship between TSS concen- Mg mg/L 27.7 ± 0.3 45.6 ± 4.2 tration and zeta potential over the range of 0.1-8.0 g TSS/L [31]. Na mg/L 2,916 ± 79 691 ± 38 In our analysis, the range of TSS concentrations tested for zeta potential was 0.032-1.7 g/L. Morgan et al. [32] indicated that sludge - solids carry a negative charge regardless of the sludge type. growth was observed at 140 mg N-NO3 /L (N/P = 1.33). Under - Moreover, they postulated that the amount of negative charge constant nitrogen concentration of 140 mg N-NO3 /L, and N/P ratio in the 3-26.4 range, the highest biomass concentration was observed carried by the sludge solids was due to the extracellular polymers 3- (ECPs) yield; as the activated sludge solids had higher negative at 47 mg P-PO4 /L initial P concentration (N/P = 3). They also concluded that low N/P ratio is necessary for complete nitrogen charge and ECPs yield compared to the AD sludge. The lower removal. For example, at 140 mg N/L, they found that the N/P ECPs generated in the AD sludge can be related to the potential ratio should be less than or equal to 3 for complete nitrogen removal degradation of these biopolymers by bacteria to form methane from the artificial wastewater. In contrast, phosphorus removal and carbon dioxide. On the other hand, there was a strong correla- was independent of N/P ratio [14]. Finally, source A contained tion between the dilution% and conductance in both sources; higher Fe, K, and Na concentrations compared to source B, whereas which relates to the total dissolved solids concentration. the concentrations of Ca and Mg were higher in source B. 3.3. Filtration Using Polyester Filter Bags Although filtration using 0.45 μm syringe filters reduces turbid- ity and removes suspended solids from the diluted digestate, it Elevated concentrations of COD and organic carbon in particular does not necessarily improve the media to support the growth may negatively impact the growth of microalgae due to rapid micro- of microalgae. For example, the elevated ammonia concentrations bial growth [33]. For instance, Travieso et al. [34] reported that in filtered compared to unfiltered digestate may have negative COD concentrations as high as 1,100 mg COD/L inhibited the impact on the growth of many microalgae species. Besides, 0.45 growth of Chlorella vulgaris when cultivated using piggery wastewater. μm filtration would be expensive on a larger scale. Filtration using polyester filter bags is a convenient and relatively

a b

Fig. 1. Zeta potential (●) and conductance (■) values for the diluted AD (average ± SD; n = 3).

268 Environmental Engineering Research 21(3) 265-275

low cost method for the removal of solids from liquid media such in comparison to nitrates and organic nitrogen. Overall, filtration as wastewater. There was a significant reduction in all parameters using 10 μm filter bags is an attractive option to reduce the sus- when comparing the unfiltered diluted digestate with the filtered pended solids and COD contents of the diluted AD. However, liquid (Table 2). OD 750, which reflects the turbidity of the liquid, elevated ammonia concentrations have to be taken into consid- decreased by 0.398 units and 0.514 units for the 10 μm filtrate eration when preparing nutrient media for the growth of microalgae. for sources A and B diluted digestates, respectively. Further reduc- tions were not significant with finer filtration using the 5 and 3.4. Hydrogen Peroxide Treatment 1 μm filter bags. For the cultivation of microalgae using the AD, The purpose of hydrogen peroxide treatment was to evaluate the the initial turbidity of the nutrient media can become a limiting potential of oxidizing the COD and ammonia in the diluted AD. factor due to light attenuation in the highly concentrated solutions. Chemical oxidation is a process that has many applications in the TSS results follow the same pattern; as 89% and 85% of the TSS water and wastewater treatment. Taste and odor control, disinfection, were retained on the 10 μm filter bags for the diluted sources hydrogen sulfide removal, and color removal are amongst the applica- A and B digestates, respectively. COD was reduced by 61% and tions of chemical oxidation in water treatment [25]. 76% when the diluted sources A and B digestates were filtered The theoretical dosage of 30% H2O2 was estimated based on through the 10 μm filter bags, respectively. Filtration using smaller the COD content. Then 0.5, 1.0, and 1.5 times the theoretical mesh sizes did not result in significant additional reductions in dosages were applied for each source as indicated in the Materials COD. Total nitrogen was reduced by 27% and 39% when the and Methods section. As shown in Fig. 3, TSS concentrations diluted sources A and B digestates were filtered through the 10 decreased as a result of hydrogen peroxide pretreatment. The max- μm filter bags, respectively. Finer filtration did not yield any addi- imum reduction observed in the case of source A was 29% with tional reductions in nitrogen concentrations. The reduction in 1.5 × dosage, while the maximum reduction was 13% for source phosphorus was more apparent by filtering the diluted digestate B with the same dosage. Similar pattern can be noticed with OD through the 10 μm filter bags; as 55% and 68% of the total phospho- 750 as the initial OD 750 value decreased from 0.383 with no rus was removed from sources A and B diluted digestates, hydrogen peroxide to 0.293 with 1.5 × dosage for source A, whereas respectively. Similar to total nitrogen, finer filtration did not yield for source B, OD 750 decreased from an initial value of 0.556 any additional significant reduction in the total phosphorus content to 0.484 when the 1.5 × dosage was applied. There was a slight in both digestates. decrease in the COD (< 5%) when half the theoretical dosage The impact of filtration on the N/P ratio was higher in the of hydrogen peroxide was used for both sources. However, for case of source B digestate as the ratio increased from 4.6 to 8.5, higher dosages (1 and 1.5 × dosage), the COD values were higher whereas the initial ratio in source A digestate was 4 which increased than the initial, which suggests that the residual hydrogen peroxide to 6.5 after filtration as shown in Fig. 2. In addition, the remainder interfered with the spectrophotometric COD measurement. of the nitrogen in the filtrate was 80-90% ammonia; which may Similar trends were observed when the digestate samples were not be the favorable nitrogen form for many microalgae species treated with a combination of UV and hydrogen peroxide. This may indicate that hydrogen peroxide is not an effective oxidant for the COD in the diluted AD. The same interference was observed with the total nitrogen measurement, particularly for source B (Fig. 4). On the other hand, ammonia concentrations decreased with an increase in the hydrogen peroxide dosage. The maximum reduction was 22% for source A with 1.0 dosage + UV whereas 1.0 × dosage without UV resulted in 15% reduction in ammonia. The maximum reduction of ammonia for source B was 16% when the 1.5 × dosage of hydrogen peroxide was added to the solution. It is not clear however, whether the ammonia that is lost has been converted to nitrogen gas or to nitrate, as the total nitrogen Fig. 2. N/P comparison between the unfiltered diluted AD and the tests were inconclusive due to the interference of hydrogen peroxide 10 μm filtrate for sources A and B (average ± SD; n = 3).3.4. Hydrogen residual. Peroxide Treatment

Table 2. Characterization of the Unfiltered and 10, 5, and 1 μm Filtered Diluted AD (average ± SD; n = 3) Filter Mesh Size OD 750 TSS (mg/L) COD (mg/L) N (mg N/L) P (mg P/L) (μm) A 1% B 2% A 1% B 2% A 1% B 2% A 1% B 2% A 1% B 2% Unfiltered 0.520 ± 0.011 0.675 ± 0.014 417 ± 8 530 ± 4 553 ± 5 627 ± 12 55.7 ± 0.9 57.3 ± 1.3 13.8 ± 0.4 12.5 ± 0.2 10 0.122 ± 0.005 0.161 ± 0.008 46 ± 2 78 ± 2 214 ± 9 149 ± 4 40.5 ± 0.3 35.0 ± 0.1 6.2 ± 0.1 4.1 ± 0.1 5 0.113 ± 0.002 0.157 ± 0.002 43 ± 1 75 ± 2 207 ± 3 146 ± 7 40.4 ± 0.7 34.5 ± 0.2 6.0 ± 0.2 4.0 ± 0.1 1 0.098 ± 0.001 0.146 ± 0.004 31 ± 1 59 ± 1 186 ± 3 142 ± 2 39.5 ± 0.5 34.5 ± 0.3 6.0 ± 0.1 4.0 ± 0.1 A 1%: 100-fold diluted AD source A B 2%: 50-fold diluted AD source B

269 Husam A. Abu Hajar et al.

In summary, hydrogen peroxide pretreatment was not an effec- 3.5. Supernatant Characterization tive way to decrease the COD of the diluted AD or to oxidize This experiment was conducted based on a previous observation the organic nitrogen and ammonia. This can be related to the that solids tend to settle over time in the diluted AD, unlike the high alkalinity in the AD which decreases the efficiency of advanced raw digestate in which solids remain in suspension. It is thought oxidation processes such as ozone or peroxide; as bicarbonate that the dilution process helps in improving the settleability of is a radical scavenger [20]. the digestate. Pere et al. [29] indicated that the heavily loaded sludges tend to have higher zeta potential; i.e., they were more hydrophilic, which could be due to the increased content of ECPs, which affects the viscosity as well as the settling characteristics due to bioflocculation [29, 32]. The optical density values presented in Fig. 5 indicate that the more concentrated dilutions exhibited the highest reductions of OD 750 as the overall OD 750 reductions were 1.91, 1.29, and 0.76 for the 5%, 3.5%, and 2% source A dilutions, respectively and 2.25, 1.66, and 1.00 for the 10%, 7%, and 4% source B dilutions, respectively. This is equivalent to 83-87% reduction in TSS concen- tration for both sources; as the final TSS concentrations after 3 h were 320, 191, and 81 mg/L for the 5%, 3.5%, and 2% source A dilutions, respectively and 352, 235, and 120 mg/L for the 10%, 7%, and 4% source B dilutions, respectively. However, the high Fig. 3. TSS and COD concentrations for various hydrogen peroxide OD 750 values observed in the supernatant after 3 h of settling dosages (average ± SD; n = 3). may not be suitable for the growth of microalgae due to the higher turbidity and thus the lower light penetration, suggesting additional dilution might be required. Fig. 6 and 7 show the COD, nitrogen, and phosphorus concen- trations in the supernatant as a function of time. As indicated in the Materials and Methods section, supernatant samples were

extracted every 45 min and analyzed, and the C/Co ratios were plotted for each parameter over time; where Co is the initial concen- tration for the well-mixed sample, while C is the measured concen- tration at time t. It is clear from Fig. 6 that the major reduction in COD occurred within the first hour for both sources. Regardless of dilution, COD decreased by approximately 55-60% for source A and 65-70% for source B in the first hour. The overall COD reductions after 3 h of monitoring were 64-82% for source A and 74-78% for source B. Approximately 30% of total nitrogen was lost in the first hour in source A due to solids settling whereas the reduction in source B was in the range of 27-52% (Fig. 7a, Fig. 4. Nitrogen and ammonia concentrations for various hydrogen per- 7b). The reduction in phosphorus was more apparent than nitrogen oxide dosages (average ± SD; n = 3).

a b

Fig. 5. TSS and OD 750 values in the supernatant as a function of time (average ± SD; n = 4).

270 Environmental Engineering Research 21(3) 265-275

a b

Fig. 6. COD concentration in the supernatant as a function of settling time with respect to the initial COD concentration at time 0 (average ± SD; n = 4).

a b

c d

Fig. 7. Nutrients concentrations in the supernatant as a function of settling time with respect to the initial concentrations at time 0 (a) total N source A, (b) total N source B, (c) total P source A, and (d) total P source B (average ± SD; n = 4). as approximately 70 - 80% of source A’s total phosphorus was to those achieved with polyester filter bags; however, the cost lost in the first hour compared to 70-73% for source B (Fig. 7c, of filters decreases the feasibility of filtration in comparison to 7d). As a result, the N/P ratio varied over time as shown in Fig. 8. supernatant extraction. Additionally, the supernatant experiments In summary, the supernatant extraction method appears to be can be useful in evaluating the settling per unit depth and/or the most attractive and least expensive pretreatment method. area over time. For example, if we consider the 5% dilution of Compared to hydrogen peroxide treatment, supernatant extraction source A, the COD removal in the first 45 min was 491 mg/L, was more effective in decreasing the TSS and COD contents of from which the COD flux for the control volume of 30 mL and the digestate. On the other hand, the removal efficiencies of TSS over the depth of 5.5 cm is equivalent to 3.57 mg/cm/h. Therefore, and COD of the supernatant extraction method were comparable if the same removal efficiency is desirable and the initial COD

271 Husam A. Abu Hajar et al.

a b

Fig. 8. N/P ratio in the supernatant as a function of settling time (average ± SD; n = 4). content is known, the required retention time can be estimated resembled each other considerably, but it is always important based on the reactor depth assuming a discrete settling where to characterize the digestate prior to microalgae cultivation; as solids do not tend to flocculate and particles settle without inter- the anaerobic digestion process and its digestate can vary sig- action [35]. nificantly depending on the feedstock and the operational In general, after diluting the two sources, both behaved similarly conditions. in terms of nutrients as well as the suspended solids removal. A range of 5-250 mg N/L was targeted by extracting the super- However, nitrogen decreased to lower levels in source B compared natant of the 5.71% diluted digestate which yielded a nitrogen to source A. This means that in order to match a certain nitrogen concentration of 250 mg N/L and sequentially diluting this super- concentration, the supernatant extracted from source B has to natant as shown in Table 3. The 0.45 μm filtered digestate was be diluted even less than source A. As a result, the turbidity diluted as well to a range of 0.2 - 10% (Table 3). of the target dilution will be higher, which will affect the photo- As shown in Fig. 9, the nutrient media contributed to the initial trophic growth of microalgae due to light attenuation. Moreover, turbidity of some cultures, mainly the 5.71%, 4.57%, and 2.29% source B dilutions appeared to have separation in the form of supernatant, as the OD 750 measurements of these dilutions foamy layer on the surface, which interfered with the process (without microalgae inoculation) were 0.780, 0.619, and 0.285, of supernatant extraction and in fact contributed to more solids respectively. Shortly after the cultures were inoculated, OD 750 in the decanted liquid. Accordingly, source A was selected as of these three dilutions in addition to the 1.14% dilution cultures a nutrient source for the growth of the microalga N. oleoabundans. decreased, but after 4-6 days, OD 750 started to increase again. This may indicate a lag phase of the microalgal growth concurrent 3.6. Microalgae Cultivation with direct consumption of the particulate substrate or bacterial The microalga N. oleoabundans was cultivated using diluted AD production of extracellular enzymes to hydrolyze and solubilize as stated earlier. Source A AD was used, however, it was a different the particulates. In order to verify this, additional vials were in- batch from the earlier one characterized in the previous sections. oculated with 4 mL of the supernatant dilutions and 1 mL deionized For instance, the total nitrogen in the well-mixed digestate was water. The OD 750 of the new vials was monitored in order to 4,775 mg N/L compared to the initial batch value of 5,567 mg assess if the particulate matter dissolution was a result of direct N/L. Total phosphorus in this batch was 1,507 mg P/L compared to 1,381 mg P/L in the initial batch. Similarly, 0.45 μm syringe filtered nitrogen and phosphorus concentrations were 2,525 mg P/L and 483 mg P/L, respectively. Finally, the supernatant, which was extracted after 3 h of gravity settling, had undiluted nitrogen, phosphorus, and ammonia concentrations of 4,460 mg N/L, 538 mg P/L, and 2,490 mg N/L respectively. These concentrations were actually close to those obtained in the supernatant extraction ex- periment discussed earlier, where the majority of nitrogen remained in suspension regardless of the dilution. For instance, the average undiluted nitrogen concentration for the last 2 h of the supernatant extraction experiment was 4,633 mg N/L for the 5% dilution, 4,386 mg N/L for the 3.5% dilution, and 4,229 mg N/L for the 2% dilution. The corresponding undiluted phosphorus concentrations averaged 517 mg P/L for the 5% dilution, 510 mg P/L for the 3.5% dilution, and 511 mg P/L for the 2% dilution. As a result, the two batches Fig. 9. Supernatant AD as a nutrient medium (average ± SD; n = 5).

272 Environmental Engineering Research 21(3) 265-275

Table 3. Target Nitrogen and Phosphorus Concentrations in the Diluted AD Supernatant and Filtrate (average ± SD; n = 3) Target Nitrogen Concentration Supernatant Filtered (mg N/L) Dilution % Total P (mg P/L) Dilution % Total P (mg P/L) 250 ± 13.2 5.71 30.7 ± 0.69 10 48.3 ± 0.52 200 ± 8.2 4.57 24.6 ± 0.54 8 38.6 ± 0.42 100 ± 4.4 2.29 12.3 ± 0.27 4 19.3 ± 0.21 50 ± 2.2 1.14 6.1 ± 0.14 2 9.7 ± 0.11 25 ± 1.1 0.57 3.1 ± 0.07 1 4.8 ± 0.05 10 ± 0.4 0.23 1.2 ± 0.03 0.4 1.9 ± 0.02 5 ± 0.2 0.11 0.6 ± 0.01 0.2 1.0 ± 0.01 consumption by microalgae or another reason such as bacterial activity. The results indicated that after 5 days, the OD 750 of the 5.71%, 4.57%, 2.29%, and 1.14% decreased by 0.072, 0.045, 0.017, and 0008, respectively (Fig. 11). As a result, it is likely that both native bacteria and microalgae contribute to the decrease or dissolution of the particulate matter. The growth in the lower dilutions resembled that in the BG-11 culture in terms of continuous increase in optical density, although the growth provided by BG-11 was significantly higher than the 0.11-0.57% dilutions (P < 0.05). After 16 days of monitoring, the OD 750 values for the 2.29-5.71% dilutions were not significantly different (P < 0.05) and the three curves were approaching the same point, regardless of the significant differences in the initial OD 750 readings amongst these three dilutions. However, after 36 days of monitoring, the 2.29% dilution had the highest OD Fig. 10. Filtered AD as a nutrient medium (average ± SD; n = 5). 750, which was significantly higher than any other dilution (supernatant or filtered) except the 4.57% supernatant. Considering the lower initial OD 750 reading, the 2.29% may be considered the optimum dilution for the supernatant, despite the insignificant differences between the 2.29% and 4.57% curves by the end of the monitoring period. For the filtered digestate cultures, the 4% dilution achieved the highest biomass concentration expressed as OD 750 as of day 16 followed by BG-11, 2%, and 1% dilutions, respectively (Fig. 10). In summary, it appears that a nitrogen concentration of 100 mg N/L (7.3 mM) provided the optimum growth for both categories (supernatant and filtered). However, the higher optical density observed in the supernatant dilution may be due to the initial turbidity of the supernatant media compared to the filtered media. Nitrogen form may have contributed to the differences among the two categories; since nitrogen exists entirely as ammonia in Fig. 11. Supernatant AD (4 mL) and DI water (1 mL) (average ± SD; the filtered digestate, whereas both ammonia and organic nitrogen n = 5). exist in the supernatant. Li et al. [6] indicated that nitrate is the favorable nitrogen form to the microalga N. oleoabundans followed that high ammonium concentrations may cause growth inhibition, by urea and ammonium. Moreover, they compared 3, 5, 10, 15, but the toxic effects could be reduced by following a fed-batch and 20 mM nitrate concentrations and concluded that the optimum or continuous flow patterns; thus, reducing the initial inhibition nitrate concentration for cell growth was 10 mM while 5 mM [3]. On the other hand, Tam and Wong [17] indicated that microalgae was optimum for high lipid production. They also suggested that in general prefer ammonium over nitrate or organic nitrogen, espe- 15 mM or higher may inhibit cell growth [6]. Even though our cially under continuous cultivation conditions. Tam and Wong testing configuration differs from that used by Li et al. [6], the [19] revealed that 20-250 mg N/L ammonia concentrations did suggested optimum dilution is well within the range of 5-10 mM not inhibit the growth of Chlorella vulgaris. Additionally, pH affects for optimum biomass and lipid productivities. Levine et al. [3] the inhibition due to ammonia elevated concentrations. At pH obtained similar results when they observed a significantly better values less than 8, nitrogen exists mostly in the non-toxic ammo- growth with nitrate compared to ammonium. In fact, they indicated nium form as opposed to the toxic ammonia which exists under

273 Husam A. Abu Hajar et al. alkaline conditions [19]. Park et al. [20] reported that the microalga 4. Conclusions Scenedesmus accuminatus experienced inhibition at ammonium concentration up to 100 mg NH -N/L. 200 mg NH -N/L or higher 4 4 In this study, several pretreatment methods were applied to the resulted in decreasing the final biomass concentration, but the anaerobic digestate from two sources. These pretreatment methods inhibition became severe with an increase to 1000 mg NH -N/L 4 included hydrogen peroxide oxidation, filtration, and supernatant [20]. Wang and Lan [14] found that the microalga N. oleoabundans extraction. It was found that diluting the digestate and allowing consumed ammonium faster than nitrate, and they hypothesized it to settle for a certain time resulted in decreasing the COD as that ammonium is the preferred nitrogen form for this microalga. well as the turbidity in the supernatant. However, N/P ratio in- In our experiment, the highest ammonia concentration was 249 creased in the supernatant; as the reduction of phosphorus was mg N/L for the filtered 10% dilution. At this dilution, the biomass considerably higher than nitrogen. Finally, the microalga N. oleoa- concentration as expressed by OD 750 was relatively low in the bundans was cultivated using diluted anaerobic digestate super- early stages, and in fact it was the lowest after around 23 d. After natant and filtrate. It was found that the supernatant provided that, the growth increased and towards the end of the monitoring slightly better growth compared to the filtered media, which is period, the OD 750 of the 10% filtered exceeded the 1%, 0.4%, thought to be due to the high ammonia concentrations in the and 0.2% filtered cultures. As a result, there was no clear sign filtered digestate. Moreover, the 2.29% diluted supernatant, which of ammonia inhibition at the concentrations tested; however, the is equivalent to a total nitrogen concentration of 100 mg N/L, growth at high ammonia concentrations appeared to be slower appeared to provide the optimum growth of the microalga N. in the early stages. oleoabundans. It was attempted to scale up the microalgae culti- There were several attempts to scale-up the cultivation of the vation to a raceway pond, but the culture was invaded by other microalga N. oleoabundans in a 100-L raceway pond by using microalgae species. Several attempts were made in order to elimi- the 2.29% AD supernatant as a nutrient medium. However, it nate the contamination and maintain a unialgal culture, but the has been observed that the culture was not purely N. oleoabundans, extended cultivation time allowed other species to prevail. as Scenedesmus sp. and cyanobacteria cells were identified within In general, the combination of anaerobic digestion and micro- the culture. Filtering of the invasive species was attempted to algae cultivation is an attractive solution for biofuels production. maintain a unialgal culture using filter bags with opening sizes The former produces biogas while decreasing the COD of the waste, of 10 and 5 μm. This technique was effective in removing cyanobac- and generates digestate that is rich with nutrients. This digestate teria cells; however, Scenedesmus sp. cells were still dominant is widely used as a fertilizer but it was proven that it can support in the 5 μm filtrate. This was attempted several times; however, the growth of microalgae. Moreover, combining the two processes due to the relatively long cultivation period, other species grew has other promising potentials such as scrubbing the carbon dioxide in addition to N. oleoabundans. from the biogas using the microalgae and the anaerobic digestion Closed photobioreactors generally have the advantage of less of the microalgae biomass to produce biogas. contamination risks and easier contamination control when com- pared to open systems [36]. While widely used for the cultivation of high oil content microalgae, open ponds susceptibility to con- Acknowledgments taminants invasion such as bacteria, viruses, and other algae may limit their applicability on a commercial scale [36, 37]. Some This work was funded entirely by the National Science Foundation of the practices suggested in the literature to mitigate microalgal (NSF) through the Sustainable Energy Pathways (SEP) program culture contamination include growing species that have been (Award # 1230961). Dr. Husam is currently working for the University identified with less contamination risk when cultivated outdoor of Jordan. even for longer periods due to their tolerance to extreme conditions such as high alkalinity or salinity. These species include Dunaliella, Chlorella, Spirulina, and Arthrospira [37, 38]. Menetrez [37] in- References dicated that genetically modified algae may become a method to limit contamination. Moreover, allowing native invasive species 1. Gouveia L, Marques A, Da Silva T, Reis A. Neochloris oleoa- that are acclimated to the local conditions to take over and grow bundans UTEX #1185: A suitable renewable lipid source for instead can reduce the contamination concerns [38, 39]. Consequently, biofuel production. J. Ind. Microbiol. Biotechnol. 2009;36: closed photobioreactors might be more suitable for the cultivation 821-826. of the microalga N. oleoabundans. Future growth experiments 2. Gouveia L, Oliveira A. Microalgae as a raw material for biofuels in the raceway ponds will employ Scenedesmus sp. under shorter production. J. Ind. Microbiol. Biotechnol. 2009;36:269-274. cultivation periods as maintaining a unialgal culture in the open 3. Levine RB, Costanza-Robinson MS, Spatafora GA. Neochloris system for multiple weeks is extremely difficult. This will reduce oleoabundans grown on anaerobically digested dairy manure the likelihood that other invasive species such as cyanobacteria for concomitant nutrient removal and biodiesel feedstock will have time to dominate the ecosystem [36, 38]. Finally, subject- production. Biomass Bioenerg. 2009;35:40-49. ing the culture to extreme conditions in temperature, light, or 4. Pruvost J, Van Vooren G, Cogne G, Legrand J. Investigation pH may favor the growth of the target microalgae strain if it is of biomass and lipids production with Neochloris oleoabundans tolerant to these extreme conditions [36]. in photobioreactor. Bioresour. Technol. 2009;100:5988-5995.

274 Environmental Engineering Research 21(3) 265-275

5. Gour RS, Kant A, Chauhan RS. Screening of micro algae for and oil production on agricultural anaerobic waste effluents. growth and lipid accumulation properties. J. Algal. Biomass Bioresour. Technol. 2011;102:5076-5082. Utln. 2014;5:38-46. 23. Sturm BS, Lamer SL. An energy evaluation of coupling nutrient 6. Li Y, Horsman M, Wang B, Wu N, Lan C. Effects of nitrogen removal from wastewater with algal biomass production. Appl. sources on cell growth and lipid accumulation of green alga Energ. 2011;88:3499-3506. Neochloris oleoabundans. Appl. Microbiol. Biotechnol. 2008;81: 24. Clesceri L, Greenberg A, Eaton A. Standard Methods for the 629-636. Examination of Water and Wastewater, 20th ed. Washington 7. Meti BS, Sailaja B. Treatment of sugar process waste water D.C: American Public Health Association; 1998. and biogas production using algal biomass. Int. J. Eng. Res. 25. Crittenden JC, Trussell RR, Hand DW, Howe KJ, Tchobanoglous Technol. 2014;3:61-67. G. Water treatment principles and design. Hoboken, NJ: John 8. James SC, Boriah V. Modeling algae growth in an open-channel Wiley & Sons; 2005. raceway. J. Comput. Biol. 2010;17:895-906. 26. Ksibi M. Chemical oxidation with hydrogen peroxide for domes- 9. Terry KT, Raymond LP. System design for the autotrophic tic wastewater treatment. Chem. Eng. J. 2006;119:161-165. production of microalgae. Enzyme Microb. Technol. 1985;7: 27. James DE. Culturing algae. Carolina Biological Supply 474-487. Company; 1978. 10. Santos AM, Lamers PP, Janssen M, Wijffels RH. Biomass and 28. Forster CF. The surface of activated sludge particles in relation lipid productivity of Neochloris oleoabundans under alka- to their settling characteristics. Water Res. 1968;2:767-776. line-saline conditions. Algal Res. 2013;2:204-211. 29. Pere J, Alen R, Viikari L, Eriksson L. Characterization and 11. Da Silva TL, Reis A, Medeiros R, Oliveira AC, Gouveia L. dewatering of activated sludge from the pulp and paper Oil production towards biofuel from autotrophic microalgae industry. Water Sci. Technol. 1993;28:193-201. semicontinuous cultivations monitorized by flow cytometry. 30. Liao BQ, Allen DG, Droppo IG, Leppard GG, Liss SN. Surface Appl. Biochem. Biotech. 2009;159:568-578. properties of sludge and their role in bioflocculation and 12. Giovanardi M, Baldisserotto C, Ferroni L, Longoni P, Cella settleability. Water Res. 2001;35:339-350. R, Pancaldi S. Growth and lipid synthesis promotion in mixo- 31. Su B, Qu Z, Song Y, Jia L, Zhu J. Investigation of measurement trophic Neochloris oleoabundans () cultivated with methods and characterization of zeta potential for aerobic gran- glucose. Protoplasma 2014;251:115-125. ular sludge. J. Environ. Chem. Eng. 2014;2:1142-1147. 13. Urreta I, Ikaran Z, Janices I, et al. Revalorization of Neochloris 32. Morgan JW, Forster CF, Evison L. A comparative study of the oleoabundans biomass as source of biodiesel by concurrent nature of biopolymers extracted from anaerobic and activated production of lipids and carotenoids. Algal Res. 2014;5:16-22. sludges. Water Res. 1990;24:743-750. 14. Wang B, Lan CQ. Biomass production and nitrogen and phos- 33. Roudsari FP, Mehrnia MR, Asadi A, Moayedi Z, Ranjbar R. phorus removal by the green alga Neochloris oleoabundans Effect of microalgae/activated sludge ratio on cooperative treat- in simulated wastewater and secondary municipal wastewater ment of anaerobic effluent of municipal wastewater. Appl. effluent. Bioresour. Technol. 2011;102:639-644. Biochem. Biotech. 2014;172:131-140. 15. Yang Y, Mininberg B, Tarbet A, Weathers P. At high temperature 34. Travieso L, Benítez F, Sánchez E, Borja R, Martín A, lipid production in Ettlia oleoabundans occurs before nitrate Colmenarejo MF. Batch mixed culture of Chlorella vulgaris depletion. Appl. Microbiol. Biot. 2013;97:2263-2273. using settled and diluted piggery waste. Ecol. Eng. 2006;28: 16. Choi HJ, Lee SM. Effects of microalgae on the removal of nu- 158-165. trients from wastewater: Various concentrations of Chlorella 35. Carlsson B. An introduction to sedimentation theory in waste- vulgaris. Environ. Eng. Res. 2012;17:S3-S8. water treatment. Systems and Control Group, Uppsala 17. Tam NFY, Wong YS. The comparison of growth and nutrient University; 1998. removal efficiency of Chlorella pyrenoidosa in settled and acti- 36. Mata TM, Martins AA, Caetano NS. Microalgae for biodiesel vated sewages. Environ. Pollut. 1990;65:93-108. production and other applications: A review. Renew. Sust. 18. Park K, Lim B, Lee K. Growth of microalgae in diluted process Energ. Rev. 2010;14:217-232. water of the animal wastewater treatment plant. Water Sci. 37. Menetrez MY. An overview of algae biofuel production and Technol. 2009;59:2111-2116. potential environmental impact. Environ. Sci. Technol. 19. Tam NFY, Wong YS. Effect of ammonia concentrations on 2012;46:7073-7085. growth of Chlorella vulgaris and nitrogen removal from media. 38. Rodolfi L, Zittelli GC, Bassi N, et al. Microalgae for oil: Strain Bioresour. Technol. 1996;57:45-50. selection, induction of lipid synthesis and outdoor mass culti- 20. Park J, Jin HF, Lim BR, Park KY, Lee K. Ammonia removal vation in a low-cost photobioreactor. Biotechnol. Bioeng. from anaerobic digestion effluent of livestock waste using green 2009;102:100-112. alga Scenedesmus sp. Bioresour. Technol. 2010;101:8649-8657. 39. Sheehan J, Dunahay T, Benemann J, Roessler P. A look back 21. Franchino M, Comino E, Bona F, Riggio VA. Growth of three at the U.S. Department of Energy’s Aquatic Species Program microalgae strains and nutrient removal from an agro-zootech- - Biodiesel from algae. U.S. Report NREL/TP-580-24190. nical digestate. Chemosphere 2013;92:738-744. National Renewable Energy Laboratory: Golden CO; 1998. 22. Yang Y, Xu J, Vail D, Weathers P. Ettlia oleoabundans growth

275